Please reread my post. My use of "perfect" was in terms of the effects -- as in, a 100% reduction of methamphetamine's effects, such that it'd be reasonable to call it a "perfect" drug. Not in terms of safety, nor rolling die for people's lives.

Though I suppose you'd be "playing statistics on with people's lives" on a drug that has a 75% effect as opposed to a 100% (ie, perfect) effect. Since the drug isn't as effective as expected, treatment as a whole isn't as effective, so deaths will happen. A statistically better outcome overall, but still a "statistical game" as you call it.